4.2
Model development and forcing
The model setup used here is similar to the one described by Cougnon et al. [2013], and in Chapter 2, using the same horizontal and vertical grid. In this version, some modifications have been made to improve the vertical mixing scheme (B. Galton-Fenzi personal communication). Previously, the strong mixing induced by the intense latent heat polynya in winter, convected the entire water column in one time step. In this version of the model, the maximum convection depth of the surface oceanic boundary layer is kept to a minimum value in order to avoid mixing the entire water column in one time step, and is done using a maximum vertical mixing coefficient of 0.4 m2 s−1. In addition, one year in the model is equal to
364 days, this is to allow tidal forcing to be periodic in a year. Whereas previously in Chapter 2 realistic tidal predictions were used, following the same methods as in [Galton-Fenzi et al. 2012]. The tidal forcing used in the current chapter is detailed later in this section.
The bathymetry (Figure 4.1a) and ice draft (see Figure B.1 for more details) used in the pre-calving simulation (PRE) are the same as described in Chapter 2 [Cougnon et al. 2013], with the exception that the ice drafts of the Ninnis icebergs have been updated. The smaller Ninnis iceberg locked in the fast ice south of the B9B iceberg now has an ice draft of 500 m depth, and the larger Ninnis iceberg east of the MGT has an ice draft of 450 m (B. Legr´esy personal communication). Previously a 200 m ice draft was used without changing the bathymetry underneath. The B9B iceberg has an ice draft of 300 m and is located between two relatively shallow banks, the Ninnis bank (∼200 m) and a shallow area southeast of the iceberg (∼200 m). The ice drafts of the three main ice shelves in the domain are shown in Figure 4.1, for the MGT (panel b), the Ninnis Ice Shelf (panel c) and the Cook Ice Shelf (panel d). Also, a minimum water column thickness of 20 m is kept beneath each ice shelf (and iceberg) to allow basal melt rate, so the iceberg in the model are not physically grounded but act as they were.
The bathymetry and ice draft used in the post-calving simulation (POST) also now have an improved iceberg configuration relative to Chapter 2. The position of the B9B and Ninnis icebergs have been determined using satellite imagery (mainly AQUA MODIS images provided by NASA) from the austral summer 2012-2013 [Lieser et al. 2013]. The images show that the B9B iceberg moved from east of the MGT to settle north of Commonwealth Bay (west of the MGT, see Figure
4.2. MODEL DEVELOPMENT AND FORCING 86
4.1a), during the austral summer of 2011 [Lacarra et al. 2014]. The smaller Ninnis iceberg, previously located in the fast ice north of the Ninnis Ice Shelf, has been located near the new Mertz Glacier front since the calving event occurred. The larger Ninnis iceberg broke apart and a smaller piece is now grounded near the Ad´elie sill.
A climatology of ‘permanent’ fast ice (between 2010 and 2012) updated from Fraser et al. [2012], is also included in the model. The fast ice in the model is set to have an ice draft of 10 m, except for the fast ice southeast of the MGT, which is set with an ice draft of 35 m. However, the icebergs and fast ice in the region have changed location since the simulations were run, and their current spatial distribution di- verges from the geometry used in the POST simulation presented in this chapter. In December 2014, the small Ninnis iceberg moved from the Mertz ice front to the west of the Ad´elie depression, south of the other Ninnis iceberg and north of the B9B iceberg. This movement formed a barrier to the westward advection of sea-ice. Also, some fast ice southeast of the MGT broke apart and moved out of the domain. Subsequently in spring 2015, the B9B showed further signs of break-up.
4.2. MODEL DEVELOPMENT AND FORCING 87
Figure 4.1: Bathymetry (in m) of the model (a) overlaid with the model ice mask contour (grey contours). Solid grey outline corresponds to the ice mask for the PRE simulation and the solid white outline corresponds to the ice mask for the POST simulation. The light grey contours are the bathymetry contours every 500 m on the continental shelf (until 1500 m) and every 1000 m for the deeper part of the model domain. Notable features are indicated on the bathymetry. In the ocean: AB: Ad´elie Bank; AD: Ad´elie Depression; AS: Ad´elie Sill; MB: Mertz Bank; MD: Mertz Depression; MS: Mertz Sill; NT: Ninnis Trough; DT: D’Urville Trough. Along the continent: Watt Bay (WB), Commonwealth Bay (CB), and Dumont D’Urville base (DDU). Bottom panels are the ice draft (in m) for the 3 main ice shelves in the domain, Mertz Ice Tongue (b), Ninnis (c) and Cook (d). Dashed line contours on b, c and d show the 300 m, 600 m and 900 m ice draft contours, the light grey outline is the ice mask post-calving. The dark grey line across the MGT cavity on panel b, shows the transect described in Section 4.4.
4.2. MODEL DEVELOPMENT AND FORCING 88
using the year 2009 for the PRE simulation and 2012 for the POST simulation. Without a dynamic sea ice model, the fine-scale polynya activity is resolved by forcing the surface of the model with monthly heat and salt fluxes. These fluxes are based on sea ice concentration from a climatology derived model using Special Sensor Microwave Imager data (SSM/I – Tamura et al. [2011; 2016]) that is adjusted to a 364 day-year. Some modifications were needed to apply these air/sea forcing. In the previous simulation [Cougnon et al. 2013], at the onset of freezing conditions, the surface of the ocean was set to cool down and instantly become saltier at the same time (see Appendix A for more details). In addition, during summer the surface of the ocean was warming at temperatures higher than expected (up to approximately +8◦C). For the current simulations, the surface of the ocean is cooled to approximately the surface freezing temperature before to apply the salt flux at the surface of the ocean to simulate the brine rejection from the sea ice formation. During summer, the surface temperature is maintained to 0 ◦C maximum, which is a reasonable limit on the continental shelf according to the summer observations available in this region [Lacarra et al. 2011]. More details about these issues and the modifications are given in Appendix A.
The choice of the year 2009 for the PRE simulation forcing was made after analysing the monthly heat and salt fluxes that were averaged over the Mertz Glacier Polynya (MGP) area for the period 1992 to 2013 (see Figure B.3 for more details). The period from 2007 to 2009 was identified as a sustained period of relatively strong polynya activity with a winter average (May to September inclusive) of ∼ -164 W m−2,
while the average over the pre-calving period (1992-2009) was -159 ± 17 W m−2.
Similarly, the salt flux average for 2007 to 2009 was∼0.82 kg m−2, while the average
for 1992 to 2009 was 0.82± 0.1 kg m−2. As a result, 2007 to 2009 is considered as
being a representative period for the pre-calving MGP region. Ultimately 2009, the year prior to the calving event, was chosen and a single year forcing was preferable to a pre-calving climatology, when compared to a single year forcing for the post- calving simulation restricted to one year forcing due to forcing availability. In the post-calving scenario, the year 2012 was chosen because of clear visibility of the permanent features situated in the region between 2010 and 2012 (A. Fraser personal communication). In summary, the results from these simulations are not restricted to the year chosen for the forcing, they can be compared with other years of similar salt and heat flux intensity.
4.2. MODEL DEVELOPMENT AND FORCING 89
monthly wind record from ERA-interim product [Dee et al. 2011], that is adjusted to a 364 day-year for 2009 and 2012. The same lateral boundary forcing is used in both PRE and POST simulations. Lateral boundary fields, including potential temperature, salinity and horizontal velocities are relaxed to a climatology that is calculated from monthly fields from ECCO2 for the period 1992-2013 [Menemenlis et al. 2008; Wunsch et al. 2009] and are adjusted to 364 day-year. The model includes analytic tidal forcing at the lateral boundaries. The frequencies of the four main tidal constituents are adjusted to be periodic in 14 days, following the same design as Pingree and Griffiths [1981a;b]. A periodic tidal signal of 26 cycles for a 364 day-year in the model is implemented to facilitate the analyses. It is important to note that salinity values used in the model are on the Practical Salinity Scale (PSS78) and are dimensionless. The total run time of the model simulation was 33 years for each simulation. This 33 years run includes a spinup phase of 30 years to reach equilibrium, using a repeating loop of the climatology forcing. The last three years of the run are used for the analyses.